Seed recommendation system

Abstract

Aspects of the disclosure relate to an algorithmic- or machine learning-based iterative seed product recommendation system. A seed recommendation platform may analyze seed genetics data, environmental data, seed similarity parameters, fitness parameters, etc., to generate an initial seed product recommendation. Based on user feedback, the seed recommendation platform may iteratively update various modules used for generating seed product recommendation and provide revised seed product recommendations.

Description

SEED RECOMMENDATION SYSTEMTECHNICAL FIELD

[0001] Aspects of the disclosure relate to a computer-implemented system for providing and implementing seed recommendations.BACKGROUND

[0002] A yield associated with a seed product is dependent on multiple factors. For example, yield associated with a seed product may be based on genotype and environmental factors, among other parameters. Availabilities of thousands of varieties of seed products, each with an associated genotype, along with effects of possible variations in growth environment makes providing a recommendation for a particular field / environment a challenging task.SUMMARY

[0003] Aspects of the disclosure provide scalable technical solutions that address and overcome the problems associated with generating and / or modifying seed product recommendations for a specific field. Various examples herein describe an iterative, algorithmic or specifically machine learning-based approach that applies user-generated feedback to provide and / or update seed product recommendations.

[0004] A computer-implemented method for optimizing crop yield in a grower field, as described herein, may comprise a feedback-loop adjusted seed recommendation engine. A computing device, associated with the seed recommendation engine, may receive agricultural data from a plurality of data repositories comprising a genetic sequence data source, environmental data source, and a crop management data source. The computing device may generate a farm-specific seed portfolio based on a plurality of parameters (e.g., environmental data associated with the grower field, yield risk factors, seed characteristics, etc ). The computing device may further rank available seed products in the portfolio; a relative yield margin model (RYMM) may be employed in this ranking. The ranking may, for example, account for genotype of the seed products and the environmental data associated with the grower field. The computing device may generate, for the available seed products, corresponding seeding rate prescriptions. The seeding rate prescriptions may be uniform or variable. The computing device may determine and send, based on the rankingof available seed products, a preliminary seed product recommendation. The preliminary seed product recommendation may comprise an indication of a first set of seed products and corresponding first seeding rate prescriptions, which may be uniform or variable. The computing device may receive, from the user device, feedback associated with the preliminary seed product recommendation. The feedback may comprise at least one of: updated environmental data associated with the grower field, updated crop management data associated with the grower field, an indication of seed product availability, or an indication of one or more seed products. The computing device may generate, based on the received feedback associated with the preliminary seed product recommendation, a revised seed product recommendation. The revised seed product recommendation may comprise an indication of a second set of seed products and corresponding second seeding rate prescriptions, which may be uniform or variable. The computing device may send the revised seed product recommendation. The revised seed product recommendations may be applied by one or more farm devices. For example, based on receiving the revised seed product recommendation, the one or more farm devices may plant at least one seed product, of the second set of seeding products, in the grower field in accordance with a corresponding second seeding rate prescription.

[0005] These features, along with many others, are discussed in greater detail below.BRIEF DESCRIPTION OF THE DRAWINGS

[0006] The present disclosure is illustrated by way of example and not limited in the accompanying figures in which like reference numerals indicate similar elements and in which:

[0007] FIG. 1 shows an example operation of a seed recommendation platform.

[0008] FIG. 2 shows an example determination of seed similarity indexes.

[0009] FIG. 3 shows an example determination of fitness scores.

[0010] FIG. 4 shows an example application of a trained machine learning model to compare predicted performance of two seed product genotypes in view of environmental data associated with a location of interest.

[0011] FIG. 5 shows an example representation the machine learning framework for predicting a relative yield advantage (RY A).

[0012] FIG. 6 shows an exemplary variable seeding rate logic as applied for determination of seeding rate recommendations.

[0013] FIG. 7 shows an example method for iterative seed product recommendations based on feedback.

[0014] FIG. 8 shows an illustrative computing environment for providing and / or iteratively revising seed product recommendations.DETAILED DESCRIPTION

[0015] In the following description of various illustrative embodiments, reference is made to the accompanying drawings, which form a part hereof, and in which is shown, by way of illustration, various embodiments in which aspects of the disclosure may be practiced. It is to be understood that other embodiments may be utilized, and structural and functional modifications may be made, without departing from the scope of the present disclosure. It is noted that various connections between elements are discussed in the following description. It is noted that these connections are general and, unless specified otherwise, may be direct or indirect, wired or wireless, and that the specification is not intended to be limiting in this respect.

[0016] Various examples describe herein enable generation of seed product recommendations based on an iterative model that includes feedback from a user. The user may be an on-field expert who may consult with the grower to refine the recommendations as generated by a seed recommendation platform and provide feedback to the seed recommendation platform. The seed recommendation platform may process the feedback to provide final recommendations to the grower, devices / platforms associated with the grower, and / or platforms associated with seed product inventory systems. The recommendations may include indications of one or more seed product genotypes and uniform or variable rate seeding protocols / scripts associated with the seed product genotypes.

[0017] FIG. 1 shows an example operation of a seed recommendation platform 100. FIG. 1 further shows interactions between the seed recommendation platform and a user device to iteratively generate and / or refine seed product recommendations as generated by the seed recommendation platform 100. Multiple data repositories and algorithms or machine learningmodels may be used, at the seed recommendation platform 100, to generate seed product recommendations, as further described herein.

[0018] At step 104, the seed recommendation platform 100 may determine a seed product portfolio based on determined seed similarity indexes characterizing a similarity (e.g., in properties) of available seed products with seed products previously associated with / procured by a specific farm or end-user. For example, a customer metadata repository 136 may include information regarding prior procurements associated with a customer / grower. The seed recommendation platform 100 may generate seed similarity indexes associated with each available seed products (e.g., as listed in a seed inventory repository 144) with respect to seed products preferred and previously procured by the customer. The seed similarity indexes may consider disease ratings, suitability ratings, and / or other agronomic data associated with the seed products to determine seed similarity indexes.

[0019] The customer metadata repository 136 may include additional information relating to customer preferences. For example, the customer metadata repository 136 may comprise an indication of a crop-type preference, a crop-rotation preference, seed-maturity and / or trait-score preference, grain marketing preference, irrigation management preference, soil fertility management preference, and / or disease management preference. The seed recommendation platform 100 may determine a seed product portfolio based on seed products that satisfy one or more of the preferences indicated in the customer metadata repository 136.

[0020] FIG. 2 shows an example determination of seed similarity indexes. Customer procurement history may be determined on the basis of a location of interest 216 for which a seed recommendation is to be provided. For example, a user device may be used to input, to the seed recommendation engine, customer data for generating a seed recommendation. The customer data may be mapped to / indicate a location of interest 216 associated with the customer. The location of interest 216 may correspond to a field for which seed product recommendations are to be generated by the seed recommendation platform 100.

[0021] The seed recommendation platform 100 may determine a prior years’ seed product procurement data (e.g., using the customer metadata repository 136) associated with a territory comprising the location of interest 216. The seed recommendation platform 100 may further determine a trait distribution 204 associated with the previously-procured seed products. The traits204 may correspond to disease ratings, suitability ratings (e.g., with respect to the location of interest 216), and / or ratings associated with other seed characteristics of the previously-procured seed products.

[0022] The seed recommendation platform 100 may additionally determine seed product traits 208 associated with available seed products supported by the seed recommendation platform 100. For example, the seed recommendation platform 100 may determine disease ratings, suitability ratings, and / or ratings associated with other seed characteristics for the available seed products. The available seed products may be based on querying a seed inventory repository 144 associated with the seed recommendation platform 100, and different seed products may be listed as available in the repository 144 based on the region in question.

[0023] Based on the traits 208 of the available seed products and the traits 204 of previously- procured seed products, seed similarity indexes 220 may be determined for each of the available seed products. For example, a higher index may be assigned to a seed product that has similar traits to the previously-procured seed products. Determining a seed similarity index may comprise performing a vector comparison of the ratings associated with previously-procured seed products with the ratings associated with the available seed products. The ratings may be associated with pest resilience, resilience to environmental agronomic pressures, fungal resilience, disease resilience, agronomic traits, price, and / or other characteristics. Example characteristics that may be used for determining seed similarity indexes / performing vector comparison may comprise: drought tolerance, root strength, stalk strength, plant height, ratings characterizing resilience against one or more diseases, etc.

[0024] Seed similarity indexes may additionally be based on current and historical regional disease and / or agronomic pressures 212, and the corresponding expected impact of those pressures on yield outcomes. A higher seed similarity index may be assigned to seed products that have similar disease and / or pest resilience characteristics (e.g., disease ratings) to the previously- procured seed products. A database storing regional disease and agronomic pressures 212 may be used to determine specific diseases, agronomic pressures, and / or other risk factors in the location of interest 212 and determine respective similarity ratings (e.g., as associated with the diseases, agronomic pressures, and / or other risk factors) for the seed products. For example, the regional disease and agronomic pressures database 212 may comprise pest-related data. For example, thepest-related data may comprise pest species, pest load, pest genetic strain (e.g. haplotype-based), pest effect (historical / predicted) on agronomics of a crop product (e.g., yield), and / or pest control measures (e.g., comprising at least one of crop protection agent identification, standard crop protection agent spray recommendations, crop protection agent effectiveness, and / or crop protection agent cost).

[0025] The portfolio of seed products may be determined based on the seed similarity indexes. For example, the portfolio of seed products may comprise seed products that are associated with seed similarity indexes above a threshold value. The portfolio of seed products may comprise seed products which have seed similarity indexes that are above a particular threshold of seed similarity indexes of all available seed products. The seed similarity indexes may also be combined with factors such as the difference in relative seed product maturity, the predicted yield advantage compared to previously purchased products, the trait segment or technology, the seed product age, and product cost, prior to determining the recommended seed product portfolio.

[0026] At step 108, the seed recommendation platform 100 may determine, from the seed product portfolio, a set of seed products based on fitness scores of the seed products. Fitness scores of the seed product may be determined based on the location of interest 216 (e.g., a field) for which seed product recommendations are to be generated by the seed recommendation platform 100. The fitness scores may provide a granular measure of a suitability of a seed product for the field / location of interest 216.

[0027] FIG. 3 shows an example determination of fitness scores. The location of interest 216 may be used to determine localized environment features 312 associated with the field (e.g., as indicated by the location of interest 216). For example, the localized environment features 312 may be queried from an environmental data repository 124 which may include environmental data for a territory covered and / or served by the seed recommendation platform 100, and / or from a field metadata repository 140 which may include field specific information. The environmental data repository 124 may comprise geospatial data corresponding to weather, climate, remotely sensed crop and / or soil attribute data, etc., that provides a representation of environment at regional, field, and sub-filed scale. The field metadata repository 140 may comprise information relating to soil properties / types, irrigation characteristics, drainage attributes, historical / average yield productivity for specific crop types, water-response, drought tolerance characteristics, landscape positioning, product performance data, agronomic risk factors and disease risk factors, and / ormanagement practices to mitigate those risk factors. Examples include: fungal pressure, where a “no risk" value would be listed for that field, where a neighboring field that isn't sprayed would still have a risk value; and stalk lodging, where there may be a regional risk, but a lower population in the field would eliminate that risk so a “no risk" value would be listed.

[0028] Other information that may be stored in and / or determined based on the environment data repository 124 and / or the field metadata repository 140 is further described with respect to step 112, and FIGS. 4 and 5. The location of interest 216 may, as such, be associated with a vector that defines the environment features 312 as derived from the environmental data repository 124 and / or the field metadata repository 140.

[0029] A seed product may be associated with traits (e.g., characteristics) that may define its suitability when used in fields associated with specific environment features. These traits and associated suitability may be stored in a seed product traits and suitability ratings database 308 and used for determination of fitness scores. The seed product traits and suitability ratings database 308 may store, for each seed product, a corresponding suitability rating in view of one or more possible environmental features that the seed product may be used in. Suitability ratings may be determined based on observed historical performance of the seed products and / or other criteria.

[0030] For example, a crop associated with a first seed product may be associated with a strong flood tolerance. Accordingly, the first seed product may have high suitability ratings in fields prone to flooding (e.g., poor drainage). Similarly, a crop associated with a second seed product may be associated with a requirement of frequent irrigation. Accordingly, the second seed product may have high suitability ratings in fields that are well irrigated and / or are located in regions receiving frequent rainfall, and poor suitability in fields that are poorly irrigated and / or are located in regions receiving scanty rainfall.

[0031] A seed product may be associated with traits that may define its suitability for use in areas that are susceptible to a specific disease and / or pest. Accordingly, suitability ratings may also be determined for the seed products based on tolerance to various diseases and / or pests at the location of interest 216. For example, higher suitability rating may be assigned for a seed product that is more tolerant to a particular diseases and / or pests that are prevalent in the location of interest 216. The database storing regional disease and agronomic pressures 212 may be used to determine-1-specific diseases, agronomic pressures, and / or other risk factors in the location of interest 212 and determine respective suitability ratings for the seed products.

[0032] Agronomic traits, disease traits, and / or productivity traits that may be used for suitability ratings may be determined from a seed genetics repository 120. Multiple suitability ratings may be determined for each seed product in the seed product portfolio. Each suitability rating may define a suitability of the seed product for a particular parameter (e.g., environment feature, disease, agronomic pressure, etc.) associated with the location of interest 216. A fitness score 316, for each seed product, may be generated based on the multiple suitability ratings for that seed product.

[0033] The set of seed products, from the portfolio of seed products as determined at step 104, may be determined based on fitness scores. For example, the set of seed products may comprise seed products that are associated with fitness scores above a threshold value. The set of seed products may comprise seed products which have fitness scores that are above a particular threshold of fitness scores of seed products in the portfolio of seed products. The set of seed products may correspond to seed products determined to be most suitable for the location of interest 216, after considering seed product traits, environmental features, disease prevalence, and / or agronomic pressures. However, the final recommended set of seed products may not include the nominally top ranked product, based on the farm specific portfolio recommendation and expected management plan, for example where the highest ranked product is not available due to being allocated to a different set of fields, in which case the recommendation will incorporate lower ranked products while optimizing the full farm field plan.

[0034] At step 112, the seed recommendation platform 100 may determine, for each seed product in the set of seed products, a correspond yield based on one or more algorithms or machine learning models. For example, a machine learning model may be trained to learn a relationship of a seed product genotype and environment on yield performance. The machine learning model may be used to predict yields for different seed product genotypes in view of environmental features / locati on-specific environmental data associated with a location of interest (e.g., the location of interest 216).

[0035] The algorithm or machine learning model may be used to determine a relative yield advantage (RY A) measure of a seed product genotype over another seed product genotype. TheRYA may correspond to a probability of one of at least two candidate seed product genotypes outperforming another one of the at least two candidate seed product genotypes with respect to a yield at the location of interest. The RYA may be a difference between a predicted yield for one or more first candidate seed product genotypes at the location of interest as compared to a predicted yield f for one or more candidate seed product genotypes at the location of interest.

[0036] The machine learning model may be trained to learn interactions between genotype and environment for multiple different seed product genotypes and locations (e.g. from one or more training datasets). The machine learning model may be trained to learn a relationship of genotype by environment interactions with yields for a plurality of seed product genotypes. For example, the machine learning model may be trained to learn how a seed product’s yield relates to the seed product genotype and an environment that the seed product is used in. The trained machine learning model may be capable of predicting performance (e.g., yield) for one or more seed product genotypes at one or more given locations. The trained machine learning model may predict whether one or more seed product genotypes will perform better or worse at one or more given locations when compared to one or more other seed product genotypes.

[0037] The training datasets may comprise data representations of seed product genotypes and location-specific environmental data associated with fields wherein a performance (e.g., yield) of the seed product genotypes was observed. In some aspects, the genotypic data (e.g., genetic sequence data) may comprise information about the genome of a given plant or plants associated with the seed products. The genotypic data may be accessed, by the seed recommendation platform 100, from a seed genetics repository 120. The seed genetics repository 120 may comprise information corresponding to genetic markers, gene-expression, metabolites, phenotypic, agronomic traits, disease traits, productivity traits, etc.

[0038] The genotypic data may comprise, for example, a collection of genotypic markers, such as genome-wide markers, a specific subset of genotypic markers, presence or absence in the genome of specific mutations, single nucleotide polymorphisms (SNPs), insertion of bases, deletion of bases, other sequence information, or any combination thereof. For example, genotypic data may comprise genome-wide marker information, genome sequence information selected from the group consisting of SNP, quantitative trait loci (QTL), ribonucleic acid (RNA)-seq, short readgenomic sequencing, marker data, long read genome sequence information, methylation status, gene expression values, indels, haplotypes, and combinations thereof.

[0039] In some examples, the genotypic data may be obtained using high density DNA arrays, PCR-based methods, including tape arrays, TaqMan assays, Restriction Fragment Length Polymorphisms (RFLPs), Target Region Amplification Polymorphisms (TRAPs), Isozyme Electrophoresis, Randomly Amplified Polymorphic DNAs (RAPDs), Arbitrarily Primed Polymerase Chain Reaction (AP-PCR), DNA Amplification Fingerprinting (DAF), Sequence Characterized Amplified Regions (SCARs), Amplified Fragment Length Polymorphisms (AFLPs), or any combinations thereof.

[0040] The genotypic data, such as marker information, may be imputed. For example, the marker information may be imputed based on a common latent representation of the underlying marker information using global and / or local variational autoencoders (VAEs).

[0041] The seed product genotypes may correspond to inbred genotypes, hybrid genotypes, varietal genotypes, genotypes from immediate and subsequent generations, offspring genotypes or progeny genotypes thereof, or any combination of one or more of the foregoing. In some examples, plants associated with the seed products may be inbred plants, hybrid plants, varieties, immediate and subsequent generations, offspring or progeny thereof, or any combination of one or more of the foregoing. Seed products associated with any monocot or dicot plant may used with the methods and systems provided herein, including but not limited to a soybean, maize, sorghum, cotton, canola, sunflower, rice, wheat, sugarcane, alfalfa tobacco, barley, cassava, peanuts, millet, oil palm, potatoes, rye, or sugar beet plant.

[0042] The location-specific environmental data may be accessed, by the seed recommendation platform 100, from the environmental data repository 124 and / or a field metadata repository 140. The location-specific environmental data may comprise information for or relating to geographical locations such as latitude and longitude information, land features such as elevation, site topography, and / or climate conditions (e.g. weather conditions, such as wind direction, wind velocity, cloud cover, humidity, relative humidity, sunrise time, sunset time, temperature, precipitation, water vapor, vapor pressure deficit, snow depth, barometric pressure, season, heat index, visibility, dew point, air quality, storms, solar radiation, etc.). Environmental data may be obtained from remotely-sensed imagery, including visual light, infrared, near-infrared,multi- spectral, and / or hyperspectral imaging bands. The imaging bands may be combined and / or processed into vegetative indices (e.g., normalized difference vegetation index (ND VI), enhanced vegetation index (EVI), weighted difference vegetation index (WDVI), or normalized difference water index (NDWI)); parameters associated with soil type, soil substrate type (e g. sand, loam, clay), parameters associated with soil conditions (e.g., aeration level, temperature, (ground) water level, soil moisture, humidity level, pH, composition such as organic matter, degree of compaction, ground soil organic carbon estimates or capacity, soil toxicities, soil nutrients, inputs or applied products such as fertilizers, herbicides, insecticides, seed treatments, seed- or soil-applied agricultural biologicals soil drainage); (crop) plant conditions (e.g. plant population density, planting date, nutrient application, plant height, evapotranspiration rate, gross primary productivity (GPP), growing and harvesting season, growth cycle, stage of development, (e.g., plant phenological stage including flowering and grain filling), green-up, dry down, and senescence, seed type, vegetation, crop variety, chemical, physical and nutritional requirements); biotic stresses (e.g., plant disease resistance level, including but not limited to Northern Leaf Blight (NLB) and Goss's Wilt (GOSWLT), plant herbicide tolerance level, physical injuries e.g. from pathogens, herbicides, storms, plant stress); abiotic stresses (e.g., early and late root lodging, stalk lodging, brittlesnap, willowing, management decisions such as row count, irrigation, irrigation location, and tillage); previous and / or subsequent crops in a crop rotation system; crop production methods, (e.g. whether grown in the open field, in a growth chamber, or in a greenhouse); disease and pest events, weeds, and / or any combinations thereof.

[0043] The location-specific environmental data may be obtained in any suitable manner or using any technique. For example, the location-specific environmental data may be obtained using imaging devices, cameras, and sensors. The data may be from any number of sources, including but not limited to an aerial sources (e.g., satellites, including high resolution satellites, airplanes, helicopters, balloons and UAV platforms), a ground-based sources (e.g., trucks, tractors, rovers, other vehicles, objects such as weather stations that are landbound, or a mobile-source such as one or more handheld or mobile devices), or sources of data that do not fall into any of the other categories.

[0044] The algorithmic or machine learning framework may incorporate representations of spatial coordinate data in addition to the seed product genotypic data and location-specific environmental data to make predictions at the plot level. The spatial coordinate data may comprisethe relative coordinates or positioning of one or more rows or plots within a subfield or field, for example, the row, column coordinates of a plot within a subfield or a field.

[0045] Based on training using the training datasets, the model architectures, model weights, and any pre- or post-processing factors, may be written and stored so that the models may be used to predict performance of candidate seed products with new (e.g., outside training data) genotypes in new (e.g., outside training data) environments, candidate seed products with genotypes in the training data in new (e.g., outside training data) environments, or candidate seed products with new (e.g., outside training data) genotypes in environments observed in the training data.

[0046] The machine learning model may be a deep learning model or a supervised learning model. The deep learning model may implement self-attention. The machine learning model may be a deep learning transformer model. The machine learning model may be established by using, as input for training, genotypic data associated with seed product genotypes, location-specific environmental information, and observed performance data (e.g., yield) associated with the seed product genotypes. In some examples, the machine learning model may additionally be established using data representations for spatial coordinates for a spatial unit (e.g. a row, a plot, a subfield, or a field).

[0047] Any suitable machine learning models may be used in the methods and systems described herein. Types of models include, without limitation, statistical models, machine learning models, and models involving deep learning, including fully-supervised, self-supervised, or semisupervised methodologies. In some aspects, the machine learning model uses attention or selfattention mechanisms. In some aspects, the machine learning model is a classification model, a regression model, a distribution model, for example, a multivariate or univariate Gaussian distribution model, or a deep learning model, such as a transformer model. In some embodiments, the machine learning model is part of an ensemble model.

[0048] Some non-limiting examples of machine learning algorithms that can be used for the machine learning model may include supervised and non-supervised machine learning algorithms, including regression algorithms (such as, for example, Ordinary Least Squares Regression or Ridge Regression), instance-based algorithms (such as, for example, Learning Vector Quantization), decision tree algorithms (such as, for example, classification and regression trees), Bayesian algorithms (such as, for example, Naive Bayes and Bayesian Neural Networks),clustering algorithms (such as, for example, k-means clustering), association rule learning algorithms (such as, for example, Apriori algorithms), artificial neural network algorithms (such as, for example, Perceptron), deep learning algorithms (such as, for example, Convolutional Neural Networks, Residual Neural Networks, or transformer based models), dimensionality reduction algorithms (such as, for example, Principal Component Analysis), ensemble algorithms (such as, for example, Random Forests or Gradient-Boosted Trees), and / or other machine learning algorithms. In some embodiments, the supervised learning model is a deep learning neural network.

[0049] FIG. 4 shows an example application 400 of a trained machine learning model to compare predicted performance of two seed product genotypes in view of environmental data associated with a location of interest. A neural network-based encoding may be implemented to encode raw data into a latent representation for enabling meaningful representation of original data into a low dimensional space, while conserving maximum information. The neural network-based encoding may preprocess raw data into latent vectors prior to input to the machine learning model. The use of neural network encoding may minimize noise.

[0050] For example, genotypic data for the two seed product genotypes (e.g., hybrid A genotypic data 404 and hybrid B genotypic data 408) may be processed using a neural networkbased encoding at a genetic encoder 416 to generate a latent representation of genotypic data. Similarly, environmental data 412 associated with the location of interest may also be processed using a neural network-based encoding at an environment encoder 420 to generate a latent representation of environmental data. A G X E predictor 424 may correspond to a machine learning model that is trained to predict whether a first seed product genotype (e.g., hybrid A) will perform better or worse at the location of interest when compared to the second seed product genotype (e.g., hybrid B). Concatenated data from the genetic encoder 416 and the environment encoder 420 may be input into the G X E predictor 424. The G X E predictor 424 may provide an RYA measure associated with hybrid A and hybrid B. The RYA may be a difference between a predicted yield for hybrid A as compared to a predicted yield for hybrid B at the location of interest.

[0051] FIG. 5 shows an example representation 500 the machine learning framework for predicting a RYA. For training the machine learning model, the seed performance observations 502 (e.g., observed yield) for one or more seed products may be recorded in a seed performance observations database 502. Genotypic data, for the one or more seed products may be derived fromthe seed genetics repository 120. Environmental data associated with the environment where the seed performance observations 502 are obtained from the may be derived from the environmental data repository 124 and / or the field metadata repository 140. The machine learning model may be trained to predict performance (e.g., yield) of the one or more seed products based on the seed performance observations 502, the genotypic data of the one or more seed products, and the environmental data of the environment where the one or more seed products were utilized. In an arrangement, the machine learning model may correspond to a RYA predictor 504.

[0052] For making predictions associated with one or more seed products (e.g., yield, RYA) using the trained machine learning model, genotypic data for the one or more seed products may be determined from the seed genetics repository 120. Further, environmental data may be determined from the environmental data repository 124 and / or the field metadata repository 140 based on the location of interest 216 for which a prediction is to be made and a seed recommendation is to be provided. The determined genotype information and the environmental data may be applied to the machine learning model (e.g., the RYA predictor 504), which may be provide a prediction of a yield associated with a seed product, or a RYA measure for a comparison between two or more seed products.

[0053] In addition to genotypic data and environmental data, a machine learning model may use other crop management-related parameters for training and prediction. The management parameters may comprise and / or may relate to a seeding rate (e.g., plants / acre), irrigation data (e.g., fully irrigated, limited irrigation, or non-irrigated), tiling data related to water management such as drainage, tillage (e.g., conventional, no-till, strip-till, etc.), previous crop (e.g., corn, soybean, wheat, etc.), planting date, desired harvest date, soil fertility parameters (e.g., Ca, K, Mg, P, Zn, pH, etc., levels), fertility management parameters (e.g., application of specific fertilizers, such as N, P, K, etc.), use of crop-rotation, use of double cropping, disease management, livestock management, cover-crops management, weed management, the overall farm-level crop plan, etc. The management-related parameters may be end-user specific and, as such, may provide a more accurate yield prediction. The crop-management related parameters, associated with an end-user, may be stored and retrieved from the farm management repository 128.

[0054] In addition to, or instead of machine learning models, plant growth associated with seed products may be modeled using various mathematical equations to predict yield and / or RYAs in view of genotypic data, environmental data, and / or management-related parameters. For example,plat growth may be modeled from the initial germination phase to a maturity phase using mathematical equations. A prediction of the yield and / or RYA for particular seed products may be determined based on the mathematical modeling.

[0055] In addition to yields and / or RYA measures, the machine learning models and / or mathematical models may be used to predict other phenotypic performance metrics for different seed products. Similar to above, these machine learning models and / or mathematical models may apply / use genotypic data, environmental data, and / or management related parameters. Other phenotypes that may be modeled and / or predicted may include adjusted gross income (AGI), yield gain, root lodging resistance, stalk lodging resistance, brittle snap resistance, ear height, grain moisture, plant height, disease resistance, pest resistance, drought tolerance, cold tolerance, heat tolerance, salt tolerance, stress tolerance, herbicide tolerance, and / or flowering time. Information relating to generation and use of machine learning models is further described in International Application No. PCT / US2023 / 068985, the contents of which are hereby incorporated by reference in their entirety.

[0056] At step 112, the seed recommendation platform 100 may further determine, from the set of seed products as determined at step 108 and based on an applied machine learning model and / or mathematical equations (e.g., as described above), one or more seed products providing the best predicted yield, RYA, and / or phenotypic performance. For example, a ranking of the seed products based on the predicted yield, RYA, and / or phenotypic performance may be determined. The one or more seed products may comprise seed products that have the best ranking (e.g., ranking above a threshold value).

[0057] At step 116, the seed recommendation platform may generate variable rate seeding script(s) for the one or more seed products as determined at step 112. Step 116 depicts the generation of a variable rate seeding script, but a uniform rate seeding script is also contemplated. The variable rate seeding script(s) may, for each of the one or more seed products, seeding rate recommendations for a location of interest (e.g., the location of interest 216). The seeding rate recommendations may indicate different seed rates for different zones / areas within the location of interest, optimizing for a maximum profitability based on expected yield within the different zones and costs associated with the seed products.

[0058] FIG. 6 shows an example variable rate seeding logic 600 as applied for determination of seeding rate recommendations. This is an exemplary seeding logic generating a variable seeding rate recommendation, but the same or similar logic could also be used to generate a uniform seeding rate recommendation. The variable rate seeding logic 620 may use yield zone data 604 associated with the location of interest, soil attribute data 608 associated with the location of interest, and / or seed-specific seeding rate curves 612. The yield zone data 604 may be determined and / or predicted, for example, based on high resolution yield maps or historical yield data. The soil attribute data 608 may correspond to data related to drainage, fertility, and / or water-stresses, and may be based on prior performed analysis / studies in the location of interest. The seed-specific seeding rate curves 612 may comprise seeding-rate curve data associated with the one or more seed products for which the variable rate seeding logic 620 is being applied. The variable rate seed logic 620 may further be based on user input data (e.g., range of seed cost, expected produce price, etc.) to determine seeding rate for maximum retums / profitability. The yield zone data 604 and / or the soil attribute data 608 may be stored and / or retrieved from the field metadata repository 140 and / or may be determined based on information stored in the environmental data repository 124.

[0059] For example, the location of interest may be categorized (e.g., soil attribute data 608 and the yield zone data 604) into high yielding zone, moderate yielding zone, and low yielding zone. The variable rate seeding logic 620 may determine different seeding rates for the high yielding zone, moderate yielding zone, and low yielding zone in a manner that maximizes profitability. In an arrangement, the variable rate seeding logic 620 may, based on seed-specific seeding rate curves, appropriately recommend high seeding rates for the high yielding zone, moderate seeding rates for the moderate yielding zone, and low seeding rates for low yielding zone. The variable rate seeding script for the location of interest may comprise an indication of the different seeding rates and associated zones for the different seeding rates.

[0060] Parameters associated with the machine learning model (e.g., model architectures, model weights, any pre- or post-processing factors, etc.) as applied at step 112 may be stored in a model repository 132 as shown in FIG. 1. In addition to the machine learning model, the model repository 132 may also store various algorithms used for determining seed similarity indexes (e.g., at step 104), fitness scores (e.g., at step 108), and uniform or variable rate seeding scripts(e.g., at step 116) may be stored in the model repository 132. As noted above, step 116 can also involve uniform rate seeding scripts.

[0061] At step 118, the seed recommendation platform 100 may send an indication of a seed product recommendation (e.g., as determined based on steps 104, 108, 112, and 116). The seed product recommendation may comprise an indication of one or more seed products (e.g., with highest fitness scores and / or predicted to provide the best yield, RYA, and / or phenotypic performance) and associated variable rate seed scripts (or alternatively, uniform rate seed scripts) for the location of interest. For example, the seed recommendation platform 100 may send the indication to one or more user devices 148 associated with a local expert for review and / or feedback. The indication of the seed product recommendations and may correspond to a preliminary recommendation for the location of interest.

[0062] Based on review of the seed product recommendation, the local expert may provide and / or update the recommendations. For example, the expert may update the recommendation based on a review of the end-user needs / preferences and based on an understanding of the local planting areas associated with the location of interest. The updated recommendation may be processed as a feedback 152 into the seed recommendation platform 100. The feedback 152 may be used to revise / update the various machine learning models and / or methodologies used for calculating various parameters (e.g., fitness scores, seed similarity indexes, and / or uniform or variable rate seeding scripts) for determining seed product recommendations.

[0063] FIG. 7 shows an example method 700 for iterative seed product recommendations based on user feedback. At step 704, the seed recommendation platform 100 may send indications of one or more seed product recommendations (e.g., indications of one or more seed products and associated seeding rate scripts), for a location of interest, to a user device (e.g., a computing device associated with a local expert).

[0064] At step 708, the seed recommendation platform 100 may receive, from the user device, user feedback. User feedback may comprise information that may not be accounted for in initial seed product recommendations as provided by the seed recommendation platform 100. The user feedback may account for any recency bias that may be important for generating seed product recommendations (and that may not have been appropriately considered by the seedrecommendation platform 100). The user feedback may include recent / updated information relating to environmental data, farm management practices associated with the location of interest, customer (e.g., end-user / farmer) profile, field metadata, etc.

[0065] For example, if the location of interest was subject to storms and / or high wind conditions during recent times, this information may not have been appropriately considered by the seed recommendation platform 100 in generating in the initial seed product recommendations. Accordingly, an initial seed product recommendation may have a poor lodging score. The local expert may send feedback indicating that a recommended seed product should have a high tolerance to lodging. For example, the expert may send an indication of a threshold root lodging score to be used for providing seed product recommendation.

[0066] Additionally or alternatively, the user feedback may include indications of a latest seed inventory, any specific seed products as recommended by the user, any specific seed products to be excluded from seed product recommendations, field-specific uniform or variable rate seeding scripts as previously applied at the location of interest and / or recommended by the user, etc. Additionally or alternatively, the user feedback may include indications associated with availability of seed products (e.g., in a physical location serving the location of interest), a cost preference associated with seed products (e.g., as preferred by a grower), etc.

[0067] At step 712, the seed recommendation platform 100 may revise seed product recommendations for the location of interest based on received user feedback. For example, the seed recommendation platform 100 may perform one or more of the steps 104, 108, 112, and / or 116 to generate revised seed product recommendations based on user feedback. The user feedback may also be used to update information stored in the various repositories (e.g., environmental data repository 124, the farm management repository 128, the field metadata repository 140, the seed inventory repository 144, etc.).

[0068] The user feedback may be used to generate updated suitability ratings for seed product recommendations (e.g., at step 108). The user feedback may also be used to update the algorithms or machine learning model used for seed product recommendations (e.g., at step 112). For example, the seed recommendation platform 100 may update model weights and any pre- or postprocessing factors in a machine learning model used for generating seed productrecommendations. The user feedback as applied to the various machine learning models and / or data repositories may also be used to generate seed product recommendations for other end-users associated with other locations of interest.

[0069] Continuing with the example where the user feedback includes information pertaining to the location of interest being subject to storms and / or high winds, the seed recommendation platform 100 may generate revised seed product recommendations comprising seed products which meet or exceed threshold root lodging scores. For example, at step 108, the seed recommendation platform may generate / account for updated root lodging scores in fitness scores used for generating seed product recommendations.

[0070] The revised seed product recommendations may be sent to the user device for further review and / or approval. The steps 704, 708, and 712 may be repeated, as needed, until a final seed product recommendation is ready for the location of interest and / or the grower. The final seed product recommendation may be sent to a computing device associated with the grower and / or to computing platforms associated with a seed inventory system (e.g., for delivery to the location of interest, following a received approval from the grower). The final seed product recommendation may also be sent to one or more farm equipment (e.g., for initiating seeding operations).

[0071] FIG. 8 shows an illustrative computing environment 800 for providing and / or iteratively revising seed product recommendations. The computing environment 800 may comprise one or more devices (e.g., computer systems, communication devices, and the like). The computing environment 800 may comprise, for example, the seed recommendation platform 100, an enterprise platform 845, one or more farm equipment 855, and / or one or more user devices 850. The one or more of the devices and / or systems, may be linked over a communication network 840.

[0072] The communication network 840 may comprise one or more private or public networks (e.g., Internet). Communication by the devices in the communication network 840 may be via any wired communication protocol(s), wireless communication protocol(s), one or more protocols corresponding to one or more layers in the Open Systems Interconnection (OSI) model (e.g., local area network (LAN) protocol, an Institution of Electrical and Electronics Engineers (IEEE) 802.11 WIFI protocol, a 3rdGeneration Partnership Project (3 GPP) cellular protocol, a hypertext transfer protocol (HTTP), etc.).

[0073] The seed recommendation platform 100 may comprise one or more computing devices and / or other computer components (e.g., processors, memories, communication interfaces) configured to perform one or more functions as described herein. The seed recommendation platform 100 may comprise one or more of processor(s) 825, transmit / receive (TX / RX) module(s) 825, memory 810, and / or the like. One or more data buses may interconnect the processor(s) 825, the TX / RX module(s) 825, and / or memory 810. The seed recommendation platform 100 may be implemented using one or more integrated circuits (ICs), software, or a combination thereof, configured to operate as discussed herein. Memory 810 may be any memory such as a randomaccess memory (RAM), a read-only memory (ROM), a flash memory, or any other electronically readable memory, and / or the like.

[0074] The processor(s) 825 may be configured to execute machine readable instructions stored in memory 810. The memory 810 may comprise (i) one or more program modules / engines (e.g., a seed recommendation engine 820) having instructions that when executed by the one or more processors cause the seed recommendation platform 100 to perform one or more functions described herein and / or (ii) one or more databases 815 that may store and / or otherwise maintain information which may be used by the one or more program modules / engines. For examples, the one or more program modules / engines may comprise instructions that when executed by the one or more processors to generate seed similarity indexes, generate fitness scores, train and / or apply algorithms or machine learning models for generating seed recommendations, and / or generate uniform or variable rate seeding scripts. The one or more databases 815 may store one or more of the repositories as described herein (e.g., the seed genetics repository 120, the environmental data repository 124, the farm management repository 128, the model repository 132, the customer metadata repository 136, the field metadata repository 140, and / or the seed inventory repository 144). The one or more program modules / engines and / or databases may be stored by and / or maintained in different memory units of the seed recommendation platform 100 and / or by different computing devices that may form and / or otherwise constitute the seed recommendation platform 100.

[0075] The enterprise platform 845 may comprise one or more computing devices and / or other computer components (e.g., processors, memories, communication interfaces). The enterprise platform 845 may be configured to host, execute, and / or otherwise provide one or more enterprise applications. For example, the enterprise platform 845 may be used to interact with, operate,and / or modify the various program modules / engines and / or databases corresponding to the seed recommendation platform 100. The enterprise platform 845 may be associated with a seed inventory system used to manage seed inventories at one or more locations and / or to process a received request for seed products.

[0076] The user device(s) 850 may correspond to personal computing devices (e.g., desktop computer, laptop computer) or mobile computing devices (e.g., smartphone, tablet). The user device(S) 850 may be used (e.g., by a local expert) to review recommendations as provided by the seed recommendation platform 100 and / or to provide feedback for iteratively updating the recom m endati on s .

[0077] The farm equipment 855 may comprise automated planting equipment that may receive a seed product indication and an associated uniform or variable rate seeding prescription. For example, the equipment 855 may automatically disperse an indicated / recommended seed product at a rate specified by the uniform or variable rate seeding prescription at a corresponding farm location. The farm equipment 855 may comprise any other type of agricultural equipment that may receive and apply any received seed product recommendations.

[0078] One or more aspects of the disclosure may be embodied in computer-usable data or computer-executable instructions, such as in one or more program modules, executed by one or more computers or other devices to perform the operations described herein. Generally, program modules include routines, programs, objects, components, data structures, and the like that perform particular tasks or implement particular abstract data types when executed by one or more processors in a computer or other data processing device. The computer-executable instructions may be stored as computer-readable instructions on a computer-readable medium such as a hard disk, optical disk, removable storage media, solid-state memory, RAM, and the like. The functionality of the program modules may be combined or distributed as desired in various embodiments. In addition, the functionality may be embodied in whole or in part in firmware or hardware equivalents, such as integrated circuits, application-specific integrated circuits (ASICs), field programmable gate arrays (FPGA), and the like. Particular data structures may be used to more effectively implement one or more aspects of the disclosure, and such data structures are contemplated to be within the scope of computer executable instructions and computer-usable data described herein.

[0079] Various aspects described herein may be embodied as a method, an apparatus, or as one or more computer-readable media storing computer-executable instructions. Accordingly, those aspects may take the form of an entirely hardware embodiment, an entirely software embodiment, an entirely firmware embodiment, or an embodiment combining software, hardware, and firmware aspects in any combination. In addition, various signals representing data or events as described herein may be transferred between a source and a destination in the form of light or electromagnetic waves traveling through signal-conducting media such as metal wires, optical fibers, or wireless transmission media (e.g., air or space). In general, the one or more computer- readable media may be and / or include one or more non-transitory computer-readable media.

[0080] As described herein, the various methods and acts may be operative across one or more computing servers and one or more networks. The functionality may be distributed in any manner, or may be located in a single computing device (e.g., a server, a client computer, and the like). For example, in alternative embodiments, one or more of the computing platforms discussed above may be combined into a single computing platform, and the various functions of each computing platform may be performed by the single computing platform. In such arrangements, any and / or all of the above-discussed communications between computing platforms may correspond to data being accessed, moved, modified, updated, and / or otherwise used by the single computing platform. Additionally or alternatively, one or more of the computing platforms discussed above may be implemented in one or more virtual machines that are provided by one or more physical computing devices. In such arrangements, the various functions of each computing platform may be performed by the one or more virtual machines, and any and / or all of the above-discussed communications between computing platforms may correspond to data being accessed, moved, modified, updated, and / or otherwise used by the one or more virtual machines.

[0081] Aspects of the disclosure have been described in terms of illustrative embodiments thereof. Numerous other embodiments, modifications, and variations within the scope and spirit of the appended claims will occur to persons of ordinary skill in the art from a review of this disclosure. For example, one or more of the steps depicted in the illustrative figures may be performed in other than the recited order, and one or more depicted steps may be optional in accordance with aspects of the disclosure.- l-

Claims

CLAIMS1. A computer-implemented method of optimizing crop yield in a grower field by deploying a feedback-loop adjusted seed recommendation engine, the method comprising: receiving, by a computing device, agricultural data from a plurality of data repositories comprising a genetic sequence data source, an environmental data source, and a crop management data source, generating, by the computing device, a farm-specific seed portfolio through an agricultural- product specific modeling algorithm that considers a plurality of parameters comprising environmental data associated with the grower field, yield risk factors, and seed characteristics; ranking, by the computing device, available seed products in the portfolio, wherein the ranking is optionally based on a relative yield margin model (RYMM) that accounts for genotype of the seed products and the environmental data associated with the grower field, generating, for the available seed products, corresponding seeding rate prescriptions optimized for increased return on investment based on planting density and expected yield return; determining, based on the ranking of available seed products, a preliminary seed product recommendation based on current seed inventory, wherein the preliminary seed product recommendation comprises an indication of a first set of seed products and corresponding first seeding rate prescriptions; transmitting, to a user device, the preliminary seed product recommendation; receiving, from the user device, feedback associated with the preliminary seed product recommendation, wherein the feedback comprises at least one of updated environmental data associated with the grower field, updated crop management data associated with the grower field, an indication of seed product availability, or an indication of one or more seed products; generating, by the computing device and based on the received feedback associated with the preliminary seed product recommendation, a revised seed product recommendation, wherein the revised seed product recommendation comprises an indication of a second set of seed products and corresponding second seeding rate prescriptions; sending, by the computing device, the revised seed product recommendation, wherein the sending the revised seed product recommendation causes planting of at least one seed product, ofthe second set of seeding products, in the grower field in accordance with a corresponding second seeding rate prescription.

2. The method of claim 1, wherein the seeding rate prescriptions are uniform or variable rate seeding prescriptions.

3. The method of claim 1, wherein the causing planting of the at least one seed product is based on a receiving, by the computing device, a selection of the at least one seed product.

4. The method of claim 1, wherein the generating the farm-specific seed portfolio comprises: generating a seed similarity score set based on seed product traits associated with seed products corresponding to prior procurements, seed product traits of the available seed products, regional crop pressure characteristics, and a location of the grower field; generating a fitness score set based on localized environmental data associated with the grower field, the seed product traits, the regional crop pressure characteristics, and the location of the grower field; and generating the farm-specific seed portfolio based on one or more of: seed products previously associated with the farm, the seed similarity score set and the fitness score set.

5. The method of claim 4, wherein the regional crop pressure characteristics comprise disease characteristics and agronomic characteristics.

6. The method of claim 1, wherein the ranking the available seed products in the portfolio comprises ranking the available seed products based on at least one of: predicted yields of the available seed products in the portfolio; a probability of a first seed product in the portfolio outperforming a second seed product in the portfolio with respect to a predicted yield; or a difference between a first predicted yield of the first seed product in the portfolio and a second predicted yield of a second seed product in the portfolio.

7. The method of claim 1, wherein the ranking the available seed products in the portfolio comprises applying an algorithm, such as a trained machine learning model, to predict aperformance of a seed product, among available seed products in the portfolio, based on a genotype of the seed product and the environmental data associated with the grower field.

8. The method of claim 7, wherein the algorithm has as further inputs crop management- related parameters, associated with the grower field, from the crop management data source.

9. The method of claim 8, wherein the crop management related parameters comprise at least one of: a seeding rate, irrigation data, tillage information, prior crop information, planting date, desired harvest date, soil fertility parameters, fertility management parameters, indication of use of crop-rotation, tiling information, farm-level crop plan or indication of use of double cropping.

10. The method of claim 1, wherein the seed characteristics correspond to at least one of disease ratings or suitability ratings.

11. The method of claim 1, wherein feedback associated with the preliminary seed product recommendation further comprises a cost preference associated with seed products.

12. The method of claim 1, wherein the indication of one or more seed products comprises an indication of one or more preferred seed products to be included in the revised seed product recommendation.

13. The method of claim 1, wherein the indication of one or more seed products comprises an indication of one or more seed products not to be included in the revised seed product recommendation.

14. An apparatus comprising: one or more processors; and memory storing instructions that, when executed by the one or more processors, cause the apparatus to: receive agricultural data from a plurality of data repositories comprising a genetic sequence data source, environmental data source, and a crop management data source, generate a farm-specific seed portfolio through an agri cultural -product specific modeling algorithm that considers a plurality of parameters comprising environmental data associated with the grower field, yield risk factors, and seed characteristics; rank available seed products in the portfolio, wherein the ranking is optionally based on a relative yield margin model (RYMM) that accounts for genotype of the seed products and the environmental data associated with the grower field, generate corresponding seeding rate prescriptions optimized for increased return on investment based on planting density and expected yield return; determine, based on the ranking of available seed products, a preliminary seed product recommendation based on current seed inventory, wherein the preliminary seed product recommendation comprises an indication of a first set of seed products and corresponding first seeding rate prescriptions; transmit, to a user device, the preliminary seed product recommendation; receive, from the user device, feedback associated with the preliminary seed product recommendation, wherein the feedback comprises at least one of: updated environmental data associated with the grower field, updated crop management data associated with the grower field, an indication of seed product availability, or an indication of one or more seed products; generate, by the computing device and based on the received feedback associated with the preliminary seed product recommendation, a revised seed product recommendation, wherein the revised seed product recommendation comprises anindication of a second set of seed products and corresponding second seeding rate prescriptions; send the revised seed product recommendation, wherein the sending the revised seed product recommendation causes planting of at least one seed product, of the second set of seeding products, in the grower field in accordance with a corresponding second seeding rate prescription.

15. The apparatus of claim 14, wherein the seeding rate prescriptions are uniform or variable rate seeding prescriptions.

16. The apparatus of claim 14, wherein the instructions, when executed, cause generating the farm-specific seed portfolio by causing: generating a seed similarity score set based on seed product traits associated with seed products corresponding to prior procurements, seed product traits of the available seed products, regional crop pressure characteristics, and a location of the grower field; generating a fitness score set based on localized environmental data associated with the grower field, the seed product traits, the regional crop pressure characteristics, and the location of the grower field; and generating the farm-specific seed portfolio based on the seed similarity score set and the fitness score set.

17. The apparatus of claim 14, wherein the instructions, when executed, cause ranking of the available seed products in the portfolio by causing ranking of the available seed products based on at least one of: predicted yields of the available seed products in the portfolio; a probability of a first seed product in the portfolio outperforming a second seed product in the portfolio with respect to a predicted yield; or a difference between a first predicted yield of the first seed product in the portfolio and a second predicted yield of a second seed product in the portfolio.

18. The apparatus of claim 14, wherein the indication of one or more seed products comprises one or more of:an indication of one or more preferred seed products to be included in the revised seed product recommendation, or an indication of one or more seed products not to be included in the revised seed product recommendation.

19. A non-transitory computer readable medium storing instructions that, when executed, cause: receiving, by a computing device, agricultural data from a plurality of data repositories comprising a genetic sequence data source, environmental data source, and a crop management data source, generating, by the computing device, a farm-specific seed portfolio through an agricultural- product specific modeling algorithm that considers a plurality of parameters comprising environmental data associated with the grower field, yield risk factors, and seed characteristics; ranking, by the computing device, available seed products in the portfolio, wherein the ranking is optionally based on a relative yield margin model (RYMM) that accounts for genotype of the seed products and the environmental data associated with the grower field, generating, for the available seed products, corresponding seeding rate prescriptions optimized for increased return on investment based on planting density and expected yield return; determining, based on the ranking of available seed products, a preliminary seed product recommendation based on current seed inventory, wherein the preliminary seed product recommendation comprises an indication of a first set of seed products and corresponding first seeding rate prescriptions; transmitting, to a user device, the preliminary seed product recommendation; receiving, from the user device, feedback associated with the preliminary seed product recommendation, wherein the feedback comprises at least one of updated environmental data associated with the grower field, updated crop management data associated with the grower field, an indication of seed product availability, or an indication of one or more seed products; generating, by the computing device and based on the received feedback associated with the preliminary seed product recommendation, a revised seed product recommendation, whereinthe revised seed product recommendation comprises an indication of a second set of seed products and corresponding second seeding rate prescriptions; sending, by the computing device, the revised seed product recommendation, wherein the sending the revised seed product recommendation causes planting of at least one seed product, of the second set of seeding products, in the grower field in accordance with a corresponding second seeding rate prescription.

20. The non-transitory computer readable medium of claim 19, wherein the instructions, when executed, cause ranking the available seed products in the portfolio by causing ranking the available seed products based on at least one of: predicted yields of the available seed products in the portfolio; a probability of a first seed product in the portfolio outperforming a second seed product in the portfolio with respect to a predicted yield; or a difference between a first predicted yield of the first seed product in the portfolio and a second predicted yield of a second seed product in the portfolio.

21. The apparatus of claim 19, wherein the indication of one or more seed products comprises one or more of: an indication of one or more preferred seed products to be included in the revised seed product recommendation, or an indication of one or more seed products not to be included in the revised seed product recommendation.

Images